fix(cache): normalize bypassed scan failures

[mldangelo-oai]

Summary

• normalize bare unsuccessful ScanResult metadata on direct cached_scan bypasses
• add regression coverage for read-failure-aware cache bypass results

Validation

• PYTHONPATH=/Users/mdangelo/.codex/worktrees/76dd/modelaudit PROMPTFOO_DISABLE_TELEMETRY=1 /Users/mdangelo/code/modelaudit/.venv/bin/pytest tests/cache/test_cache_correctness.py::test_cached_scan_normalizes_and_skips_persisting_bare_unsuccessful_scan_result tests/cache/test_cache_correctness.py::test_cached_scan_normalizes_bare_unsuccessful_result_on_direct_bypass -q
• PYTHONPATH=/Users/mdangelo/.codex/worktrees/76dd/modelaudit /Users/mdangelo/code/modelaudit/.venv/bin/ruff check modelaudit/utils/helpers/cache_decorator.py tests/cache/test_cache_correctness.py
• PYTHONPATH=/Users/mdangelo/.codex/worktrees/76dd/modelaudit /Users/mdangelo/code/modelaudit/.venv/bin/ruff format --check modelaudit/utils/helpers/cache_decorator.py tests/cache/test_cache_correctness.py
• PYTHONPATH=/Users/mdangelo/.codex/worktrees/76dd/modelaudit /Users/mdangelo/code/modelaudit/.venv/bin/mypy modelaudit/utils/helpers/cache_decorator.py tests/cache/test_cache_correctness.py

Notes

• Standard local uv run pytest is blocked before collection on this macOS ARM host by the existing tensorrt-cu13-libs placeholder-wheel install failure.

[github-actions]

Workflow run and artifacts

Performance Benchmarks

Compared 13 shared benchmarks with a regression threshold of 15%.
Status: 0 regressions, 0 improved, 13 stable, 0 new, 0 missing.
Aggregate shared-benchmark median: 4.367s -> 4.347s (-0.4%).

Workload	Benchmark	Target	Size	Files	Baseline	Current	Change	Status
warm-cache-rescan	tests/benchmarks/test_scan_benchmarks.py::test_scan_warm_cached_repository_rescan	release-candidate	547.3 KiB	32	167.36ms	155.02ms	-7.4%	stable
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_base64]	nested_base64	98 B	1	304.3us	286.6us	-5.8%	stable
direct-malicious-upload	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_direct_malicious_upload	malicious_reduce	52 B	1	225.0us	216.4us	-3.8%	stable
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_raw]	nested_raw	78 B	1	276.1us	269.6us	-2.4%	stable
padded-multi-stream-upload	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_padded_multi_stream_upload	multi_stream_padded	4.1 KiB	1	347.9us	341.9us	-1.8%	stable
chunked-upload-stream	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_chunked_upload_stream	chunked_stream	278.2 KiB	1	114.32ms	115.77ms	+1.3%	stable
nested-payload-review	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_nested_payload_review[nested_hex]	nested_hex	130 B	1	305.2us	302.1us	-1.0%	stable
clean-training-checkpoint	tests/benchmarks/test_picklescan_benchmarks.py::test_picklescan_clean_training_checkpoint	safe_large	278.2 KiB	1	111.24ms	112.03ms	+0.7%	stable
duplicate-heavy-registry	tests/benchmarks/test_scan_benchmarks.py::test_scan_duplicate_registry_snapshot	registry-snapshot	915.2 KiB	13	574.69ms	578.34ms	+0.6%	stable
rejected-basic-auth-candidates	tests/benchmarks/test_scan_benchmarks.py::test_rejected_basic_auth_candidates_scan_linearly	-	371.1 KiB	1	2.512s	2.496s	-0.6%	stable
single-checkpoint-preflight	tests/benchmarks/test_scan_benchmarks.py::test_scan_single_checkpoint_before_load	single_checkpoint.pkl	183.0 KiB	1	103.81ms	104.46ms	+0.6%	stable
mixed-model-repository	tests/benchmarks/test_scan_benchmarks.py::test_scan_release_candidate_repository	release-candidate	547.3 KiB	32	630.80ms	632.88ms	+0.3%	stable
suspicious-pickle-intake	tests/benchmarks/test_scan_benchmarks.py::test_scan_suspicious_pickle_intake	suspicious-intake	183.8 KiB	4	151.46ms	151.63ms	+0.1%	stable